1

Supplementary Information for manuscript: 1

2

Elevation in plasma tRNA fragments precede seizures in human epilepsy 3

Marion C. Hogg1,2, Rana Raoof1,3, Hany El Naggar4, Naser Monsefi1, Norman Delanty2,4, Donncha 4

F. O’Brien5, Sebastian Bauer6,7,8, Felix Rosenow6,7,8, David C. Henshall1,2 & Jochen H.M. Prehn1,2* 5

1. Department of Physiology and Medical Physics, Royal College of Surgeons In Ireland, St. 6

Stephen’s Green, Dublin D02 YN77, Ireland. 7

2. FutureNeuro Research Centre, Royal College of Surgeons In Ireland, St. Stephen’s Green, 8

Dublin D02 YN77, Ireland. 9

3. Department of Anatomy, Mosul Medical College, University of Mosul, Mosul, Iraq. 10

4. Department of Neurology, Beaumont Hospital, Beaumont, Dublin, Ireland. 11

5. Department of Neurosurgery, Beaumont Hospital, Dublin, Ireland. 12

6. Epilepsy Center Hessen, Department of Neurology, Baldingerstr, 35043, Marburg, Germany. 13

7. Epilepsy Center Frankfurt Rhine-Main, Neurocenter, Goethe-University, Schleusenweg 2-16, 14

Haus 95, 60528, Frankfurt, Germany 15

8. LOEWE Center for Personalized Translational Epilepsy Research (CePTER), Frankfurt, Germany. 16

2

Supplementary Methods 17

18

TLE patients and healthy controls 19

A 10 ml blood sample (pre-seizure) was taken on admission. A post-seizure sample was collected 20

24 h after experiencing an electro-clinical seizure documented by video-EEG monitoring. The 21

interval between pre-seizure blood sampling and seizure occurrence varied among patients 22

(median 31 hours, range 00:11-205:46), as did the number and type of seizures experienced. 32 23

non-fasting male and female healthy control volunteers (MAR, n = 16; DUB, n = 16) were 24

recruited. 25

Interictal activity analysis 26

Video-EEG recordings from a period of 18-24 hours upon arrival to the EMU were reviewed by a 27

clinical neurologist and patients were classified into three groups based on the vEEG activity: 28

rare, occasional, and frequent. tRNA fragment levels in pre-seizure samples were compared 29

between groups. 30

Plasma preparation 31

Plasma was prepared within 1 h of collection by centrifuging at 1,300 x g, for 10 min, at 4°C, and 32

stored at -80 °C. Haemolysis was assessed using a Nanodrop 2000, and samples with A414 >0.25 33

were excluded. 34

RNA Extraction 35

3

RNA was purified from plasma using the miRCURY RNA isolation kit for biofluids (Exiqon). A 36

synthetic C.elegans miRNA-39 spike-in RNA was added before purification. RNA purified from 200 37

ul plasma was eluted in 50 ul water. 38

Small RNA sequencing (RNA seq) and analysis 39

Small RNA seq (<50 nt) was performed on pooled plasma from 16 healthy controls and 16 focal 40

epilepsy patients pre and post seizure samples. RNA libraries were generated using NEBNEXT 41

library generation kit (New England Biolabs Inc.). Single ends reads were sequenced on the 42

Illumina system by Exiqon Services, Denmark. RNA seq data has been submitted to the gene 43

expression Omnibus (GSE114701). Adapter sequences were removed and reads with a quality 44

score of <20 were removed. Reads were aligned using Tophat (v 2.0.14) and Bowtie (v 2.2.5.0), 45

allowing 1 hit per read, to a custom tRNA database built from the GtRNAdb (gtrnadb.ucsc.edu). 46

Intron locations were added for 32 tRNAs, and a “CCA” tail was manually added. Reads aligning 47

to tRNAs were pooled based on their iso-acceptor type for the initial RNA seq analysis. This 48

approach was taken due to the highly similar sequence of multiple tRNAs from the same iso-49

acceptor type. Subsequently Taqman assays were designed to recognise specific tRNA 50

fragments from each iso-acceptor type that showed high abundance and high fold change. 51

The genomic origin of the specific tRNA fragments cannot be absolutely defined due to the 52

presence of multiple copies of identical tRNA genes in the genome. Reads are expressed as 53

counts per million (CPM) to correct for differences in read depth. Mature tRNA structures were 54

downloaded from GtRNAdb 2.0 (1) and tRNA fragment secondary structures were predicted 55

using the Vienna RNAfold program (2). 56

4

Taqman assays 57

Custom Small RNA Taqman assays were designed to recognise tRNA fragments 58

(ThermoScientific). The Taqman assay technology was developed to specifically amplify mature 59

miRNAs without detecting the pre- or pri-miRNA that contains identical sequence, hence this 60

technology was used here to amplify tRNA fragments without recognising the full length tRNA. 61

The stem-loop primer used in the reverse transcription step of the Taqman assay inhibits binding 62

to sequences with 3’ extensions, such as full-length tRNAs. A similar protocol has been used to 63

selectively quantify tRNA fragments previously (3). Primary hippocampal neuron samples were 64

normalised to U6, human plasma samples were normalised to C.elegans miRNA-39 spike-in. 100 65

ng (cells or tissues) or 2 ul (biofluids) RNA was used per reverse transcription reaction. 66

Quantification was performed on the Quantstudio 5 384-well PCR machine (ThermoFisher 67

Scientific) and fold-change determined using the 2-Ct method. Outliers +/- 2 standard deviations 68

from the mean were excluded. 69

Primary hippocampal neuron culture and in vitro hyperexcitation model. 70

Primary mouse hippocampal neurons were dissected from E16-E18 C57Bl/6 embryos as 71

described (4). On DIV 12 cells were incubated with 5 M FLUO-4 for 45 minutes before media 72

was replaced with experimental buffer (120 mM NaCl, 3.5 mM KCl, 0.4 mM KH2PO4, 5 mM 73

NaHCO3, 20 mM HEPES, 1.2 mM Na2SO4, 15 mM glucose, and 1.2 mM CaCl2, 1mM MgCl2). Cells 74

were transferred to the heated stage of a LSM 5 Live microscope (Zeiss) and imaged to confirm 75

spontaneous firing of neurons. Media was replaced with experimental buffer containing 0 or 1 76

mM MgCl2, and images were collected at 5Hz. Cells were incubated for 2 hours and total RNA 77

5

and media collected, importantly 2 hours in magnesium-free experimental buffer does not 78

induce neuronal cell death (5). 79

Statistical analysis 80

Statistical analysis was performed in Graphpad Prism or SPSS. Data are fold change compared to 81

control samples. Mouse hippocampal neuron experiments were analysed by two-tailed Student’s 82

t-test. Human plasma were not normally distributed therefore Kruskal-Wallis and Wilcoxon 83

Signed Rank tests were used. For all analyses a p-value of less than 0.05 was considered 84

significant. ROC analysis was performed in SPSS to determine the area under a curve (AUC) and 85

Youdens J statistic was used to identify the optimal discriminatory tRNA level. 86

Surgically resected patient tissue 87

Focal epilepsy patients who were assessed to be suitable for surgical resection were recruited at 88

the Department of Neurology, Beaumont Hospital, Dublin, Ireland. Informed consent was 89

obtained for all patients and ethical approval was obtained from the Research Ethics Committee 90

at the Royal College of Surgeons in Ireland (REC 13/75). 91

Fresh frozen tissue was mounted in OCT and sectioned on a cryostat at -22OC. 12 um sections 92

were either mounted on SuperFrost Plus slides (ThermoScientific) for histological analysis or 93

collected in Eppendorf tubes for RNA extraction in Trizol (ThermoScientific). For histological 94

analysis Nissl staining was performed. Briefly, slides were post-fixed in 4% pFa for 10 minutes and 95

washed in PBS. Slides were stained in 0.1% cresyl violet acetate solution at 65oC for 20 minutes. 96

Slides were washed extensively in water and dipped in successive ethanol solutions, 2 dips each 97

6

of 70%, 80%, 90%, 95% (plus one drop glacial acetic acid), 100%. Slides were then incubated in 98

Histoclear solution (National Diagnostics) for twice for 5 minutes and mounted in DPX solution. 99

100

101

Supplementary References 102

1. Chan PP, Lowe TM. GtRNAdb 2.0: an expanded database of transfer RNA genes identified in 103 complete and draft genomes. Nucleic Acids Res. 2016;44(D1):D184-9. 104 2. Lorenz R, Bernhart SH, Honer Zu Siederdissen C, Tafer H, Flamm C, Stadler PF, et al. ViennaRNA 105 Package 2.0. Algorithms Mol Biol. 2011;6:26. 106 3. Honda S, Loher P, Shigematsu M, Palazzo JP, Suzuki R, Imoto I, et al. Sex hormone-dependent 107 tRNA halves enhance cell proliferation in breast and prostate cancers. Proc Natl Acad Sci U S A. 108 2015;112(29):E3816-25. 109 4. Jimenez-Mateos EM, Engel T, Merino-Serrais P, McKiernan RC, Tanaka K, Mouri G, et al. Silencing 110 microRNA-134 produces neuroprotective and prolonged seizure-suppressive effects. Nat Med. 111 2012;18(7):1087-94. 112 5. Deshpande LS, Lou JK, Mian A, Blair RE, Sombati S, DeLorenzo RJ. In vitro status epilepticus but 113 not spontaneous recurrent seizures cause cell death in cultured hippocampal neurons. Epilepsy Res. 114 2007;75(2-3):171-9. 115

116

7

Group Sex Age Diagnosis AED Group Sex Age

Dublin Cohort

TLE F 18 TLE LEV, LTG C F 25 TLE F 25 LEFT TLE LEV, LTG C F 30 TLE F 28 RIGHT TLE ESLI, LEV, ZNS C F 34 TLE F 45 LEFT TLE LEV, CBZ C F 38 TLE F 64 RIGHT TLE LEV, PHE C F 38 TLE F 79 TLE (CORTICAL DYSPLASIA) ESLI , LEV, VAP C F 45 TLE M 18 LEFT TLE OXC, CLOB C M 23 TLE M 25 LEFT TLE LEV, LAC C M 24 TLE M 27 BILATERAL TLE VAP, CLOB C M 25 TLE M 35 LEFT TLE LEV, CLOB C M 31 TLE M 36 LEFT TLE LEV, LTG C M 34 TLE M 47 LEFT TLE LEV, LAC C M 37 TLE M 52 RIGHT TLE ESLI, LAC, LEV C M 41 TLE M 61 BILATERAL TLE LEV, PHY C M 46 TLE M 67 RIGHT TLE LTG, CLOB C M 51 TLE M 74 TLE TPN, LEV C M 52

Marburg Cohort

TLE F 29 LEFT TLE CBZ,LTG,TPM C F 25 TLE F 29 RIGHT temporo-parietal epilepsy LTG, LEV C F 26 TLE F 30 LEFT TLE CBZ, LEV C F 28 TLE F 33 RIGHT FTLE LTG C F 33 TLE F 34 focal epilepsy LCM, LTG C F 35 TLE F 37 RIGHT TLE n/a C F 40 TLE F 49 LEFT focal epilepsy CBZ C F 45 TLE M 18 RIGHT TLE LEV C F 46 TLE M 23 mesial LEFT TLE LEV C M 25 TLE M 34 RIGHT TLE VPA, CBZ C M 33 TLE M 35 focal epilepsy LEV, OXC, ZNS C M 34 TLE M 46 RIGHT TLE OXC, LEV C M 35 TLE M 52 LEFT TLE OXC, LEV, LCM C M 35 TLE M 52 focal epilepsy LEV, ZNS C M 48 TLE M 57 LEFT TLE or FLE TPM C M 55 TLE M 62 RIGHT TLE LEV, PGB, Clonazepam C M 65

Table 1: Patient demographics. 32 epilepsy patients and age-matched healthy controls were recruited at two independent epilepsy monitoring units in Dublin and Marburg. Age, sex, diagnosis, and current AEDs are indicated for patients, and age and sex are indicated for healthy controls.

Table 2: RNA Seq reads aligned to tRNAs. Total reads and reads aligned to tRNAs for Control, Pre-seizure and post-seizure pooled RNA seq samples. The mean Phred score was >28 for all sequences indicating they were of very high quality.

Control Pre-seizure Post-seizure

Input 30336921 48665926 32808642

Mapped 227578 524797 272138

Mapped (%) 0.80 1.10 0.80

8

A)

Human_GluCTC TCCCTGGTGGTCTAGTGGTtAGGATTCGGCGCTCTCACCGCCGCGGCCCGGGTTCGATTC 60

Mouse_GluCTC TCCCTGGTGGTCTAGTGGTtAGGATTCGGCGCTCTCACCGCCGCGGCCCGGGTTCGATTC 60

5'GluCTC TCCCTGGTGGTCTAGTGGTTAGGATT---------------------------------- 26

**************************

Human_GluCTC CCGGTCAGGGAA 72

Mouse_GluCTC CCGGTCAGGGAA 72

5'GluCTC ------------ 26

B)

Mouse_AlaTGC GGGGATGTAGCTCAGTGGTAGAGCGCATGCTTAGCATGCATGAGGtCCTGGGTTCGATCC 60

Human_AlaTGC GGGGATGTAGCTCAGTGGTAGAGCGCATGCTTAGCATGCATGAGGtCCCGGGTTCGATCC 60

5'AlaTGC GGGGATGTAGCTCAGTGGTAGAGC------------------------------------ 24

************************

Mouse_AlaTGC CCAGCATCTCCA 72

Human_AlaTGC CCAGCATCTCCA 72

5'AlaTGC ------------ 24

C)

Human_GlyGCC GCATGGGTGGTTCAGTGGTAGAATTCTCGCCTGCCACGCGGGAGGCCCGGGTTCGATTCC 60

Mouse_GlyGCC GCATGGGTGGTTCAGTGGTAGAATTCTCGCCTGCCACGCGGGAGGCCCGGGTTCGATTCC 60

5'GlyGCC GCATGGGTGGTTCAGTGGTAGAATT----------------------------------- 25

*************************

Human_GlyGCC CGGCCCATGCA 71

Mouse_GlyGCC CGGCCCATGCA 71

5'GlyGCC ----------- 25

Figure 1: Alignment of human and mouse tRNA sequences and tRNA fragments identified in this study.

tRNAs are highly conserved indicating assays designed to human tRNA fragments would also detect tRNA

fragments in mouse samples.

Figure 2: U6 levels are constant across mouse hippocampal neuron experiments. A) Intracellular and B)

Extracellular average U6 Ct values from primary mouse hippocampal neurons cultured in the presence

(Mg+) or absence (Mg-) of Magnesium show no significant difference in levels. C) Extracellular U6

normalised to C.elegans miRNA 39 spike-in also shows no significant difference between Mg+ and Mg-

cultures.

9

Table 3: Summary of ROC analysis from Figure 3.

5'GlyGCC 5'AlaTGC 5'GluCTC

AUC 0.816 0.916 0.802

p-value 0.000027 1.86E-08 0.000069

Youdens 1.36 1.33 2.13

Sensitivity 0.67 0.97 0.59

Specificity 0.93 0.87 0.9

10

Figure 3: Analysis of tRNA fragment levels in males and females. Separating the controls and patients analysed in the main manuscript Figure 3 by sex revealed that 5’AlaTGC is significantly elevated in pre-seizure males and females and significantly decreases post seizure in both groups. 5’GlyGCC and 5’GluCTC were significantly deceased following seizures in males. Data were analysed using Kruskal-Wallis test where A) 5’GlyGCC in females p=0.0129, B) 5’AlaTGC in females p=0.0006, C) 5’GluCTC in females p=0.0074, D) 5’GlyGCC in males p=0.0033, E) 5’AlaTGC in males p=0.0004, and F) 5’GluCTC in males p=0.0036. Controls include 14 female and 18 male, epilepsy patients include 13 females and 19 males.

11

Figure 4: Analysis of tRNA fragments that are not elevated in pre-seizure samples compared to controls.

Two tRNA fragments were chosen for further analysis to highlight that not all tRNA fragments are elevated

in pre-seizure samples. A) RNA seq analysis of reads aligning to tRNAs ranked by fold change between pre-

seizure and control samples indicated #34 ValAAC and #51 ProAGG were higher in controls than pre-

seizure samples. Custom Taqman assays were developed to analyse levels of B) 5’ValAAC and C) 5’ProAGG

levels in 32 pre and post seizure samples and 32 healthy controls. 5’ValAAC was not detectable in some

samples (n= 14-15 samples per group).

12

Figure 5: Analysis of tRNA fragments across 4 time points in epilepsy patients. Plasma samples collected

at intervening time points were available for 24/32 focal epilepsy patients and tRNA fragment levels were

analysed. T1 = on arrival to the EMU, T2 = 24 hours after T1 if no seizure occurred, T3 = 1 hour after

seizure, and T4 = 24 hours after seizure. Figure 3 in the main text displays data from T2 and T4 time points

for all 32 patients. One or more seizures occurred between T2 and T3 time points. All tRNA fragment levels

were significantly higher in pre-seizure samples compared to post seizure samples. Data was analysed by

Kruskal-Wallis test where * indicates p< 0.05.

13

Figure 6: Correlation of time interval between pre-seizure blood collection and onset of seizure and

tRNA fragment level. A-C) Correlation analysis for all samples, and D-F) Correlation for samples collected

within under 50 hours before seizure onset. Pre-seizure samples collection to seizure onset time interval

(hours) is plotted on the x axis and tRNA fragment level normalised to C.elegans spike-in on the y axis.

Correlation was assessed with Spearman’s r-value showing no significant correlation between time

interval and plasma tRNA fragments levels when analysing all samples; however, when restricting samples

to under 50 hours a significant correlation is observed with 5’GluCTC (panel F) and 5’GlyGCC and 5’AlaTGC

show a similar trend.

14

Figure 7: tRNA fragments are detectable in surgically resected focal epilepsy patient brain tissue. A)

5’GlyGCC, B) 5’AlaTGC, and C) 5’GluCTC were quantified in surgically resected cortical and hippocampal

tissue from five of the focal epilepsy patients in the Dublin cohort of the study. tRNA fragments were

detected in all regions analysed but no significant difference in levels could be detected. D) Histological

analysis of Nissl stained neighbouring sections revealed no gross structural changes were apparent in

these tissue samples. Scale bar 100 μm.

15

Figure 8: Analysis of structural abnormalities detected by MRI and tRNA fragment levels in pre-seizure

samples. A) Summary of MRI findings, patients were classed as MRI positive or MRI negative, and 3

patients whose results were unclear/indeterminate were excluded from the analysis. Pre-seizure levels of

B) 5’GlyGCC, C) 5’AlaTGC, and D) 5’GluCTC were increased in patients with no sign of structural

abnormalities compared to those with lesions detected by MRI, with 5’GlyGCC levels significantly different

as analysed by Mann-Whitney U test, p = 0.02. This data suggests that elevated tRNA fragment levels are

not due to underlying tissue damage or scar formation which can be detected by MRI. Abbreviations: HS:

Hippocampal Sclerosis, PNH: Periventricular Nodular Heterotopia, AVM: Arterio-Venous Malformation,

MCA: Middle Cerebral Artery, MTS: Mesial Temporal Sclerosis.

16

Figure 9: tRNA fragment levels in patients that did not have AEDs reduced and experienced seizures.

Four focal epilepsy patients from the Dublin cohort did not have their AED medication reduced upon

admittance to the EMU and experienced electro-clinical seizures. Comparing pre-seizure tRNA fragment

levels between patients with and without AED reduction, we found no significant difference (Mann-

Whitney U test).

Figure 10: tRNA fragment levels in focal epilepsy patients with reduced AEDs that did not experience

seizures. No signifcant changes in tRNA fragments A) 5’GlyGCC, B) 5’AlaTGC, or C) 5’GluCTC were detected

in plasma samples collected 24 hours apart in patients whose AED medications were reduced but did not

go on to expereience seizures. No significant difference in tRNA fragment levels was detected.

17

Figure 11: tRNA fragment levels in patients with psychogenic non-epileptic seizures (PNES). Six patients

admitted to the EMU were subsequently diagnosed with PNES, analysis of tRNA fragment levels at

timepoints T1-T4 indicated no significant difference in tRNA fragment levels were detected, and the

“seizure-like” event occurred between time points T2 and T3.

Figure 12: Analysis of pre-seizure tRNA fragments levels in relation to interictal activity. Video-EEG

recordings from a period of 18-24 hours prior to seizure onset were reviewed by a Clinical Neurologist and

patients were classified into 3 groups: Rare, Occasional, and Frequent interictal spiking activity. There was

no significant difference in levels of A) 5’GlyGCC, B) 5’AlaTGC, or C) 5’GluCTC however all showed higher

levels in patients experiencing occasional or frequent interictal activity.

18

Figure 13: Analysis of pre-seizure tRNA fragment levels according to type of seizure experienced. All

patients from the Dublin cohort (n = 16) experienced Complex Partial Seizures (CPS); however, some

patients progressed on to generalized tonic-clonic seizures (GTCS). Comparison of tRNA fragment levels

in patients with and without GTCS revealed no significant difference indicating pre-seizure plasma tRNA

fragment levels cannot discriminate seizure type.